Robinhood Markets Inc. is in the business of getting people to trade stock (and options, and cryptocurrencies) on their phones. The more people who sign up for accounts, and the more they trade, the happier Robinhood is. The way it makes money is that each time its customers trade stocks (or options, or cryptocurrencies), Robinhood sends their order to a market maker, and the market maker pays Robinhood a tiny fee for the right to execute the order. Unlike many other retail brokers, Robinhood charges these market makers a variable fee that is, roughly speaking, higher for very volatile stocks.[1] So the more people who sign up for Robinhood’s service, and the more they trade, and the more volatile the stocks that they trade are, the more money Robinhood makes.

By this standard, the last week of January 2021 was incredibly, incredibly good for Robinhood, an amazingly perfect week. That was the height of the GameStop Corp. meme-stock mania: The whole world’s attention was focused on the soaring price of GameStop stock, which was leading a lot of people to sign up for Robinhood and trade GameStop, and GameStop was very volatile and so Robinhood got paid a lot for trading it. Robinhood both increased its future value by signing up a lot of customers, and increased its present profits by extracting a lot of money from them each day.

There was, however, a problem. When Robinhood’s customers buy a stock, the trade does not settle right away: The customer agrees to buy the stock, and the market maker agrees to sell it to them, but they don’t actually exchange money for stock until two business days later. There is a credit component to the trade: The buyer has to deliver the money in two days, and the seller has to deliver the stock. If the stock shoots up over those two days, you might worry that the seller will back out of the deal; if the stock crashes, you might worry that the buyer will back out of the deal. In practice, the stock market deals with this risk through clearinghouses: One giant entity (for US stocks, it’s the National Securities Clearing Corp., a subsidiary of the Depository Trust & Clearing Corp.) guarantees all the trades, and clearinghouse members, like Robinhood, post cash with the clearinghouse to guarantee their customers’ trades.[2]

Broadly speaking, Robinhood has to post more cash with the clearinghouse as its customers trade more stocks. And it has to post more cash as those stocks are more volatile: The more likely it is that a stock will crash or moon, the bigger the credit risk is, so the more money is required. 

By this standard, the last week of January 2021 was incredibly bad for Robinhood. The GameStop meme-stock mania caused a lot of its customers to buy GameStop, and caused the price of GameStop to be very volatile, which caused Robinhood to have to post enormous amounts of cash with the clearinghouse, and it happened not to have that cash lying around. 

And so Robinhood’s otherwise great week came shockingly close to killing it.

Last week the US House Financial Services Committee released a report about that week at the end of January 2021, and it is fascinating reading. It is mostly about Robinhood, and specifically it is about the tension inside Robinhood between the fact that it was a great week and the fact that it almost blew up Robinhood.

So on the one hand, the meme-stock week was great for Robinhood’s growth, and the people whose job it was to make Robinhood grow were excited. On Tuesday, Jan. 26, Elon Musk tweeted “Gamestonk!!,” which of course pushed up interest in trading GameStop on Robinhood. A Robinhood product manager emailed about this: “FYI massive spike [in Robinhood account openings] in last 30m likely caused by Elon Musk tweet” and “we could probably interact with this movement to promote RH growth.” 

But as the week went on, Robinhood got more nervous, both about its cash needs and about whether its technology was up to all the activity. So here is a discussion — at about 11:24 p.m. on Wednesday, Jan. 27 — between a brokerage product manager at Robinhood and the firm’s head of data science:

Product manager: conflict brewing

Product manager: we have to keep the growth flywheel running

Product manager: webull is right on our tail

Head of data science: haah dont worry, we need to survive first

They wanted to grow to pull ahead of the competition, but growth that week was becoming a problem. Here is how close Robinhood came to blowing up the next day:

On the morning of January 28, 2021, Robinhood had approximately $696 million in collateral already on deposit with the NSCC, leaving it with a collateral deficit of approximately $3 billion, which it was required to post to satisfy the NSCC’s clearing fund requirement or risk being in violation of the NSCC’s rules and potentially losing the ability to clear trades for their customers altogether. [President of Robinhood’s clearing operation Jim] Swartwout confirmed that this amount came as a surprise to Robinhood and explained to Committee staff that they had anticipated and prepared for the $1.4 billion of collateral deposit requirements that represent “core” charges, but because they did not model for Excess Capital Premium charges, Robinhood therefore did not expect and had not arranged adequate funding for the additional $2.2 billion Excess Capital Premium charge. On the morning of January 28, 2021, Jim Swartwout texted [Robinhood Chief Operating Officer] Gretchen Howard at 6:29 a.m. EST, writing “Huge liquidity issue.” 

That is, the clearinghouse called Robinhood and said it needed to post about $3 billion to cover its obligations, and Robinhood didn’t have the money. Ultimately the clearinghouse waived most of the collateral requirement that day, for Robinhood and some other firms in similar predicaments. If it hadn’t (emphasis added):

Without the NSCC’s waiver of Robinhood’s Excess Capital Premium charge, Robinhood’s nonpayment would have constituted a “serious rule violation” according to the NSCC’s rules. When a clearing-broker cannot deposit the required collateral, the member is in default to the clearinghouse and NSCC may “cease to act” for that member under its rules, as the NSCC did for Lehman Brothers on September 24, 2008 and MF Global on October 31, 2011. When NSCC ceases to act, the clearinghouse assumes control of the defaulted member’s portfolio and liquidates it. This is done to limit the risk that the defaulted member poses to NSCC and to other nondefaulting NSCC members, who can be subject to mutualized losses if the collateral already held by NSCC is insufficient to cover losses on the defaulter’s portfolio. Robinhood leadership remained aware during the morning of January 28, 2021 that the NSCC could effectively eliminate the company’s ability to clear their client’s trades and liquidate the firm’s holdings.

That would have been bad! “Hypothetically what happens if a firm can’t meet their morning NSCC margin settlement,” an operations manager asked their former boss that morning in a text message. Nothing good! 

Robinhood responded to the problem — of not having enough money to support its explosive growth — by raising capital, but also by limiting customer trading in meme stocks, imposing PCO, “position closing only,” restrictions to prevent customers from buying more GameStop:

As Robinhood employees worked through Wednesday, January 27, 2021, to code position limits for meme stocks, they struggled with how to frame the trading restrictions to the public and seemed to want to avoid giving their own clients the real reasons for imposing restrictions. A product manager at Robinhood Financial asked, “Do we have a customer facing rational we can provide? In response, a manager in Robinhood’s brokerage responded, “The real reason is firm risk and us needing to control the velocity of trading. …But we shouldn’t expose that.”

It also slowed down new account approvals:

According to Robinhood operations staff, turning off auto approval meant that Robinhood was still “accepting applications, but approval is turned off. This means a large number of customers (were) still applying but (wouldn’t) be approved immediately,” delaying their ability to transact on the platform. The decision to throttle new account creation arose spontaneously on the morning of January 28, 2021. During the afternoon of January 28, 2021, Robinhood Director of Account Operations estimated that approximately 300,000 customer applications were waiting in the queue to be approved. Robinhood employees discussed when to turn auto approval back on throughout the day on January 28, 2021 but decided to keep the auto approvals turned off for the remainder of the day due to risk concerns associated with increased load on their platform. As a Robinhood employee said in a chat discussing when to resume approving accounts, “any additional load takes us to the bottom faster.”

Ultimately Robinhood’s decision to turn off auto approval had a significant impact on the number of new accounts on its platform. On the morning of January 29, 2021, the next day, Robinhood Financial’s Director of Brokerage Risk estimated that the number of customers waiting in the queue for their accounts to be approved had increased to approximately 730,000.

The meme-stock craze was so good for Robinhood that 430,000 people tried to sign up for its service overnight! And so bad for Robinhood that it had to ignore all of them. 

I don’t know. It all worked out, I guess. On Jan. 28, Robinhood Financial President David Dusseault told the Robinhood team “ah we will navigate through this nscc issue,” and “we are to big for them to actually shut us down.” Never something you want to put in writing, but not exactly wrong. The meme-stock craze made Robinhood arguably the most important company in finance for a week. It would have been better if Robinhood’s explosive success had made it rich and safe, but making it too big to fail also worked.

Russia has international bonds outstanding, and it has the dollars to make payments on them, but it has been blocked, by sanctions, from making those payments. Yesterday a deadline for making some of those payments expired, putting Russia officially into default on its bonds. Now what? The Wall Street Journal has the traditional answer:

In theory, creditors could try to seize Russian assets abroad, though it is unclear what they might go after. Some investors have suggested they might claim frozen central-bank reserves or oligarchs’ assets. Bondholders of Venezuelan debt sought assets of a state-owned oil refiner after the country’s default. In 2013, Argentina hired a private jet for the then-president’s trip to Asia and the Middle East because of the risk of creditors seizing the official aircraft. 

“I suspect, remembering what happened with Argentina, that the Americans would be keen to have creditors chasing Russian assets all over the world,” said Paul McNamara, an emerging-market fund manager at GAM. “It’s basically contracting out the job of going after Russian assets.” 

Sure. If there is some stash of dollars (or yachts, houses, etc.) that belong to Russia, the bondholders could try to seize them. I will tell you where one stash of Russian dollars might be. The Journal also reports:

Russia has plenty of money from oil and gas sales to pay its foreign debts, which are relatively small compared with the size of its economy. But allied Western governments have blocked the Kremlin’s ability to tap foreign bank accounts or use cross-border payment networks to move money.

The Treasury Department last month let a prior sanctions exemption expire that had allowed U.S. banks and investors to process and receive payments on existing Russian bonds. ...

One investor said clearinghouse Euroclear received funds for the May interest payments just before the Treasury’s exemption expiration. But the funds were frozen there because of sanctions, unable to be forwarded to his account. Lawyers say the bond documents are unclear over whether payments that reached the clearinghouse, but not the bondholder account, would constitute a formal default. 

A Euroclear spokesperson didn’t immediately respond to a request for comment.

Okay. Let’s not focus on the technical question of whether it counts as a default if the money gets to Euroclear but Euroclear does not (due to sanctions) pass it along to the bondholders. I just want to make a narrow point here, which is:

• Russia supposedly transferred the $100 million for this payment to Euroclear, where it is stuck.
• The bondholders are now, I guess, looking to seize some Russian assets abroad.
• What about that $100 million at Euroclear?

Like, Euroclear has that money,[3] and the bondholders want it, but Euroclear can’t pass it along to them because sanctions prohibit it from passing along interest payments on behalf of Russia. But if you utter the right incantations, you can turn that $100 million from “a Russian interest payment” into “Russian assets abroad that can be used to satisfy a default judgment.” And then Euroclear could give it to the bondholders. Maybe? Boy is that not legal advice.

This does not quite work as a matter of arithmetic. Russia has transferred a little bit of money to foreign banks and clearinghouses to make interest payments on its bonds, and those payments have gotten stuck, and that has led to a default, and in theory that default allows the holders of all of Russia’s billions of dollars of international bonds to accelerate those bonds and demand immediate repayment. So there could be a rush of all of the bondholders to seize money, and the $100 million that might or might not be at Euroclear won’t come close to satisfying all of their claims.

Still it is a strange sort of default. You could imagine a financial-engineering approach:

1. Get all the bondholders together to agree to accelerate the debt and try to seize Russian assets abroad.
2. The bondholders also agree to a payment waterfall: Holders of Bond A get the first $X seized, and then holders of Bond B get the next $Y, and then holders of Bonds C and D share equally in the next $Z, etc. And $X happens to be the interest payment due on Bond A next month, and $Y happens to be the amount due on repayment of Bond B at maturity next year, etc.
3. On the next interest payment date for Bond A, Russia transfers $X to a foreign account at some international bank and says “hey bondholders come seize our money.”[4]
4. The bondholders apply to the bank to seize the money and give it to the holders of Bond A under the payment waterfall.
5. The bank is like “sure whatever here you go, seizing Russian money isn’t sanctioned, in fact it is encouraged.”
6. Then Russia pays Bond B the same way.
7. Etc.

Basically you pass the bond payments along to the holders as normal, except that instead of calling them “bond payments” you call them “assets seized due to acceleration of the bonds in default.” The bondholders get paid as normal, but you get to say that Russia is getting punished for default instead of paying its debts in the ordinary course. You get to call it a default, but the bondholders get to keep getting paid.

Coinbase Global Inc. got a lot of bad press last month for describing how bankruptcy works, or might work, for cryptocurrency exchanges. “Because custodially held crypto assets may be considered to be the property of a bankruptcy estate,” Coinbase explained in a risk factor in its Form 10-Q, “in the event of a bankruptcy, the crypto assets we hold in custody on behalf of our customers could be subject to bankruptcy proceedings and such customers could be treated as our general unsecured creditors.”

This quite neutral description of a risk to Coinbase’s creditors sort of blew up, and Coinbase had to go on a weird apology tour in which they told people, well, we’re not planning to go bankrupt, and we don’t want to seize your crypto deposits. Coinbase was just warning its users, sensibly, that the law of crypto customer protection in bankruptcy is not all that clear or well developed, and that there is a serious risk that a crypto bankruptcy would not not work out the way you expect if you’re used to traditional financial-industry bankruptcies.

There is a lot of stuff like this in crypto. We talked last month about the blowup at TerraUSD, the algorithmic stablecoin. I wrote about a model in which TerraUSD was the “debt” of the Terra blockchain, while the Luna token was its “equity.” Intuitively, when it blew up, you would expect the holders of the debt to get whatever’s left (equity in the surviving company, any cash lying around, etc.), while the holders of the equity would get nothing. But in the case of Terra, things didn’t work that way: Terra launched a new blockchain, with some residual value, and gave most of it to Luna holders rather than TerraUSD holders. The thing that was supposed to be a senior claim with a fixed value turned out to be mostly worthless; the thing that was a risky bet on the growth of the network kept its value better (still not well). 

Or here is a CoinDesk story from last week about BlockFi, which announced a $250 million line of credit from FTX:

Cryptocurrency investment firm Morgan Creek Digital is attempting to raise $250 million from investors to purchase a majority stake in crypto lender BlockFi, a leaked investor call from Tuesday reveals. …

“I’ve been making calls all day,” Morgan Creek Digital managing partner Mark Yusko said on the leaked call.

According to Yusko, the FTX credit line proposal had a catch for BlockFi’s existing shareholders: It gave FTX the option to buy BlockFi “at essentially zero price.” If FTX were to exercise said option, it would effectively wipe out all of BlockFi’s existing equity shareholders, including management and employees with stock options, as well as all equity investors in the company’s previous venture rounds.

However, Yusko said on the leaked call that BlockFi founders Zac Prince and Flori Marquez had a valid reason for preliminarily accepting the terms: Of the several emergency financing offers BlockFi received, FTX’s was the only one that would not subordinate client assets to the rescuer.

In other words, unless BlockFi went with FTX, its depositors would have had to wait in line behind the new lender to be repaid.

Well they’re not “depositors” exactly, are they, not in the classic banking sense. They look like depositors, sure, but their legal rights are not so clear. 

We have talked a lot recently about how crypto has recreated the pre-2008 financial system, and is now having its own 2008 financial crisis. But this is an important difference. Traditional finance is in large part in the business of creating safe assets: You take stuff with some risk (mortgages, bank loans, whatever), you package them in a diversified and tranched way, you issue senior claims against them, and people treat those claims as so safe that they don’t have to worry about them. Money in a bank account simply is money; you don’t have to analyze your bank’s financial statements before opening a checking account. The short-term senior debt of financial institutions is “information-insensitive.” 

There is a sort of division of labor here: Ordinary people can put their money into safe places without thinking too hard about it; smart careful investors can buy equity claims on banks or other financial institutions to try to make a profit. But the careless ordinary people have priority over the smart careful people. The smart careful heavily involved people don’t get paid unless the careless ordinary people get paid first. This is a matter of law and banking regulation and the structuring of traditional finance. There are, of course, various possible problems; in 2008 it turned out that some of this information-insensitive debt was built on bad foundations and wasn’t safe. But the basic mechanics of seniority mostly work pretty well.

But they are in a sense unnatural. If you started a bank in the state of nature, and the bank had equity investors who were smart and rich and concentrated and heavily involved in the running of the bank, and then the bank took a lot of money from dispersed retail depositors who didn’t pay a lot of attention and just wanted their money back and had no real idea about the running of the bank, and then the bank ran into trouble, you might expect the equity investors to say “well we’ll take whatever money is left and stiff the depositors.” And you might not expect the depositors to be very effective at fighting back. And so the equity investors might in practice be senior to the depositors, in this state-of-nature bank. As they effectively were at Terra.

Of course if you started a bank in the state of nature nobody would give you any deposits. The reason people put their money in actual banks is that we live in a society and there are rules that protect bank deposits, and also everyone is so used to this society and those rules that they don’t think about them. Most bank depositors do not know much about bank capital and liquidity requirements, because they don’t have to; that is the point of those requirements.

Broadly speaking crypto banking (and quasi-banking) is like banking in the state of nature, with no clear rules about seniority and depositor protection. But it attracts money because people are used to regular banking. When they see a thing that looks like a bank deposit, but for crypto, they think it will work like a bank deposit. It doesn’t always.

Elsewhere here is the Financial Times on the crypto “credit crisis”:

Investors could juice their returns by taking out multiple loans against the same collateral, a process called “recursive borrowing”. This freedom to recycle capital with little restraint led investors to stack up more and more yields in different DeFi projects, earning multiple interest rates at once.

“As with the subprime crisis, it’s something really appealing in terms of yield and it looks like and is packaged like a risk-free financial product to ordinary people,” said Lennix Lai, director of financial markets at crypto exchange OKX.

The financial gymnastics left huge towers of borrowing and theoretical value teetering on top of the same underlying assets. This kept going while crypto prices sailed higher. But then inflation, aggressive interest rate rises and geopolitical shockwaves from the war in Ukraine washed across financial markets.

“It all worked during the bull run where the prices of all the assets went up only. When the prices started going down, a lot of people wanted to take their assets out,” said Marcin Miłosierny, head of market research at crypto hedge fund ARK36.

As token values plummeted, the lenders called in their loans. The process has led to the removal of more than 60 per cent, or $124bn, of the total value locked on the ethereum blockchain since mid-May in a “Great Deleveraging”, according to research firm Glassnode.

Credit Suisse, Cash and Cocaine Converge in Historic Conviction. Inside Citigroup’s Attempt to Rally Wall Street to Pressure Gun Sellers. More Hedge Funds Are Betting Against Tether as Crypto Melts Down. Trump-Tied SPAC Says New York Grand Jury Subpoenaed Its Board. BP Paid Rural Mexicans a “Pittance” for Wall Street’s Favorite Climate Solution. Pension Funds Plunge Into Riskier Bets—Just as Markets Are Struggling. Crypto Meltdown Drags Lending Returns From 25,000% to Almost Zero. How Elon Musk Helped Lift the Ceiling on C.E.O. Pay. UK Treasury takes a stake in sex party planner Killing Kittens.

If you'd like to get Money Stuff in handy email form, right in your inbox, please subscribe at this link. Or you can subscribe to Money Stuff and other great Bloomberg newsletters here. Thanks!

[1] A subplot in the House Financial Services Committee report concerns how Robinhood calculates the payment for order flow that it gets from market makers. “Most other broker-dealers calculate PFOF rebates as a flat fee per share, whereas  Robinhood calculates PFOF rebate rates based on the spread between the purchase and sale price of each security at the time of executing a trade on behalf of a customer.” At volatile times, spreads blow out, because market making is riskier: If the stock is going to move against you by $1 every time you trade, you need to charge a spread ofat least $1. Robinhood’s formula charges PFOF as, essentially, a percentage of the public-market spread. When the spread is very wide because market making is very risky, market making is not particularly profitable (you make a lot of money on spreads but lose it on adverse selection), but Robinhood *charges* its market makers more. The market makers did not love this, and there is a lot of discussion in the report about how Citadel Securities pushed back to lower its PFOF payments to Robinhood. (“An employee for Citadel Securities described Robinhood’s PFOF rebates as a runaway freight train in the days leading up to January 28, 2021.”) Robinhood got annoyed at Citadel but also proposed to cap PFOF at $0.003 per share in the meme stocks. 

[2] This cash is Robinhood’s cash, to guarantee *its* obligation, as a clearing broker, to the clearinghouse. Separately, Robinhood will want to make sure that its customers are good for their obligations to buy the stocks they agreed to buy, which it might do by, for example, requiring them to have cash in their accounts before doing trades, etc. (Not always — retail brokers offer margin trading, and sometimes let you trade before your deposits have cleared, etc. — but broadly speaking this is a solvable sort of risk management problem for the retail brokerage.) But Robinhood can’t just take the cash in its customers’ accounts and hand it over to the clearinghouse to guarantee its obligations; that cash belongs to the customers.

[3] I assume. I am not completely sure about this — the Journal reported it as a rumor, and Euroclear “didn’t immediately respond to a request for comment” — but the basic principle works even if this specific case is wrong. Russia has tried to send money to various foreign intermediaries so it can get to bondholders, and the intermediaries are subject to US and European jurisdiction and don’t pass it along.

[4] This is a hard part, I guess? The transfers here might be tricky. But, for instance, Russia is still getting hard currency by selling oil,which is allowed under the sanctions regime; presumably it can leave some of that money in a lightly guarded place and say “psst bondholders look over here.”